A knowledge dynamic mastery degree determination method, a test question recommendation method and device

Abstract

The present disclosure provides a knowledge dynamic mastery degree determination method and device, and a test question recommendation method and device. The method comprises: determining a knowledge point code of a corresponding answered test question based on multiple answered test questions of a same user; and determining a knowledge dynamic mastery degree based on the knowledge point codes of the multiple answered test questions. The answering time periods of the multiple answered test questions are different. The knowledge point code comprises knowledge point description information and an answer result of the answered test question. The multiple answered test questions contain the same knowledge point. The method provided by the present disclosure can recommend test questions with better adaptability to the user by using the knowledge dynamic mastery degree, thereby improving the learning efficiency of the user.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to a method for determining the dynamic mastery level of knowledge, a method for recommending test questions, and an apparatus. Background Technology

[0002] In recent years, the concept of online learning has gained widespread recognition in the education industry. As the number of people learning online increases, it becomes impossible for educators to accurately grasp the learning status of each user.

[0003] In related technologies, users can use online education platforms to learn, and these platforms can recommend practice questions to users based on the knowledge points they select. Summary of the Invention

[0004] According to one aspect of this disclosure, a method for determining the degree of dynamic knowledge mastery is provided, the method comprising:

[0005] A knowledge point code is determined for multiple test questions, the test questions are answered at different times, the knowledge point code includes the knowledge point description information and the answer result of the test question, and the multiple test questions contain the same knowledge point;

[0006] The user's dynamic mastery of knowledge is determined by encoding the knowledge points in multiple test questions.

[0007] According to another aspect of this disclosure, a test item recommendation method is provided, the method comprising:

[0008] The user's current level of mastery of the target knowledge point is determined based on the user's dynamic knowledge mastery information, and the user's dynamic knowledge mastery level is determined by the method described in the exemplary embodiment of this disclosure.

[0009] Based on the user's current level of mastery of the target knowledge point, multiple candidate test questions containing the target knowledge point are determined;

[0010] At least one recommended question is obtained from multiple candidate questions containing the target knowledge point, and the main knowledge point of each recommended question is the target knowledge point.

[0011] According to another aspect of this disclosure, a device for determining the dynamic level of knowledge mastery is provided, the device comprising:

[0012] The coding module is used to determine the knowledge point codes of multiple test questions answered by the same user. The multiple test questions are answered at different times. The knowledge point codes include the knowledge point description information and the answer results of the test questions. The multiple test questions contain the same knowledge points.

[0013] The determination module is used to determine the user's dynamic mastery of knowledge based on the knowledge point codes of multiple test questions.

[0014] According to another aspect of this disclosure, a test item recommendation device is provided, the device comprising:

[0015] The determination module is used to determine the user's current mastery level of the target knowledge point based on the user's dynamic mastery level of knowledge, and to determine multiple candidate test questions containing the target knowledge point based on the user's current mastery level of knowledge. The user's dynamic mastery level of knowledge is determined by the method described in the exemplary embodiment of this disclosure.

[0016] The recommendation module is used to obtain at least one recommended question from multiple candidate questions containing the target knowledge point, wherein the main knowledge point of each recommended question is the target knowledge point.

[0017] According to another aspect of this disclosure, an electronic device is provided, comprising:

[0018] Processor; and,

[0019] Memory for stored programs;

[0020] The program includes instructions that, when executed by the processor, cause the processor to perform the method according to an exemplary embodiment of the present disclosure.

[0021] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer...

[0022] A readable storage medium stores computer instructions for causing the computer to perform the method according to exemplary embodiments of the present disclosure.

[0023] One or more technical solutions provided in the exemplary embodiments of this disclosure determine the knowledge point codes of corresponding test questions based on multiple test questions answered by the same user. Since the test questions were answered at different times and contain the same knowledge points, the knowledge point codes of multiple test questions can be used to obtain dynamic change information of the knowledge point-related content. Furthermore, since the knowledge point code includes the knowledge point description information and the answer result of the test question, and the answer result reflects the user's level of mastery of the knowledge point, the user's dynamic level of knowledge mastery can be determined based on the knowledge point codes of multiple test questions. Based on this, when the method of the exemplary embodiments of this disclosure is applied to test question recommendation, it can obtain the user's current level of mastery of the target knowledge point from the user's dynamic level of knowledge mastery, then select test questions containing the target knowledge point and matching the current level of mastery from the test question bank as candidate test questions, and then filter recommended test questions from the candidate test questions. Based on this, when recommending practice questions to users, the recommended questions can be matched with the user's current mastery of the target knowledge point, ensuring that the recommended questions are compatible with the user's learning ability. This makes the recommended questions more suitable for the user's current learning situation and improves the user's learning efficiency for the target knowledge point. Furthermore, since the main knowledge point of the recommended questions is the target knowledge point, it ensures that the recommended questions primarily provide users with services to improve their abilities related to the target knowledge point. Therefore, the method of this exemplary embodiment can improve the user's learning efficiency for the target knowledge point. Attached Figure Description

[0024] The accompanying drawings, which are included to provide a further understanding of this disclosure and form part of this disclosure, illustrate exemplary embodiments of the present disclosure and are used to explain the disclosure, but do not constitute an undue limitation of the disclosure. In the drawings:

[0025] Figure 1 A schematic diagram of an example system in which the various methods described herein may be implemented according to exemplary embodiments of the present disclosure;

[0026] Figure 2 A schematic diagram of the process for determining the dynamic level of knowledge mastery, as exemplified in this disclosure, is shown.

[0027] Figure 3 A schematic flowchart of a test question recommendation method according to an exemplary embodiment of this disclosure is shown;

[0028] Figure 4 A schematic diagram illustrating the process for determining deep and complete descriptive information in an exemplary embodiment of this disclosure is shown.

[0029] Figure 5 A schematic block diagram of the functional modules of a knowledge dynamic mastery level determination device according to an exemplary embodiment of the present disclosure is shown;

[0030] Figure 6 A schematic block diagram of the functional modules of a test item recommendation device according to an exemplary embodiment of the present disclosure is shown;

[0031] Figure 7 A schematic block diagram of a chip according to an exemplary embodiment of the present disclosure is shown;

[0032] Figure 8 A structural block diagram of an exemplary electronic device that can be used to implement embodiments of the present disclosure is shown. Detailed Implementation

[0033] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0034] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0035] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below. It should be noted that the concepts of "first", "second", etc., used in this disclosure are only used to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0036] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0037] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0038] Before introducing the embodiments of this disclosure, the relevant terms involved in the embodiments of this disclosure are first defined as follows:

[0039] Deep learning is a form of machine learning, its concept originating from research on artificial neural networks. A multilayer perceptron with multiple hidden layers is a type of deep learning architecture. Deep learning discovers distributed feature representations of data by combining low-level features to form more abstract high-level representations of attribute categories or features. It is a new field in machine learning research, motivated by the desire to build and simulate neural networks that mimic the human brain's analytical learning mechanisms to interpret data.

[0040] Knowledge graphs, based on deep learning, involve constructing a "entity-relationship-entity" triple model from data information in a specific domain using deep learning algorithms, and then storing it in a graph-structured database.

[0041] Text classification refers to classifying text into one or more categories given a classification system.

[0042] The BERT (Bidirectional Encoder Representations from Transformer) model learns feature representations of the input sequence and applies these learned feature representations to different lower-level tasks. Structurally, it is a multi-layer bidirectional encoder and a pre-trained language representation model.

[0043] A fully connected neural network consists of several layers of neurons, including an input layer, an output layer, and several hidden layers. The number of neurons in each layer and the number of hidden layers are freely defined by the model builder according to the task requirements. All neurons in each layer are connected to all neurons in every other layer, while neurons in the same layer are not connected to each other.

[0044] The core idea of ​​deep fusion networks is deep fusion, which means combining the intermediate representations of several base networks as inputs to the remaining parts of each base network, and then using deep fusion on several intermediate representations.

[0045] An activation function is a function that runs on the neurons of an artificial neural network and is responsible for mapping the input of the neuron to its output.

[0046] Deep knowledge tracing models apply flexible recurrent neural networks with temporal "depth" to knowledge tracing tasks. This series of models uses a large number of artificial "neurons" to represent latent knowledge states and their temporal dynamics, allowing the learning of latent variable representations of user knowledge from data, rather than hard-coding. It solves the cold start problem and can dynamically track changes in a user's knowledge state.

[0047] Bidirectional Long Short-Term Memory (LSTM) is a type of recurrent neural network suitable for processing and predicting important events with relatively long intervals and delays in time series.

[0048] The model parameters include weight parameters and bias parameters. The weight parameters represent the slope of the hyperplane, and the bias parameters represent the intercept of the hyperplane.

[0049] A fully-connected layer is a classifier that maps the high-dimensional feature map obtained from feature extraction into a one-dimensional feature vector. This one-dimensional feature vector contains all the feature information and can be converted into the probabilities of the final classification categories.

[0050] In recent years, with the increasing popularity of online learning systems in the education environment and the growing number of online learners, educators cannot track the knowledge status of every learner and provide personalized learning guidance. As a result, learners need to search for knowledge in online learning systems from various redundant information, leading to diverse but not necessarily effective learning resources and learning paths.

[0051] The inventors analyzed online learning systems and discovered that these systems can record learners' detailed learning trajectories and answer data, reflecting their knowledge mastery. If this rich data can be utilized to conduct real-time diagnostic assessments of learners' current learning progress, and then adaptively customize recommended questions for each learner to form a learning path, learners can focus on specific knowledge points and learning patterns to achieve targeted breakthroughs, thereby improving their learning efficiency.

[0052] In related technologies, data mining of online learning systems requires extensive manual annotation of knowledge points in the questions, failing to fully utilize deep learning technology to enhance the data mining process and hindering the tracking of student status. Therefore, this exemplary embodiment provides a method for determining the dynamic level of knowledge mastery and recommending test questions. This method can mine and encode knowledge description features in test questions, and then evaluate the user's learning status through a deep knowledge tracking model to obtain the user's dynamic level of knowledge mastery, thereby completing a comprehensive assessment of the user's learning situation. Based on this, by combining the user's historical answer records to form historical test questions, and combining recommendation strategies formed by various filtering methods, customized test questions are recommended for the user, and a learning path is planned.

[0053] Figure 1 A schematic diagram of an example system in which various methods described herein can be implemented according to exemplary embodiments of this disclosure is shown. Figure 1As shown, the system 100 of the exemplary embodiments of this disclosure may include: a user device 110, a computing device 120, and a data storage system 130.

[0054] like Figure 1 As shown, the user equipment 110 can communicate with the computing device 120 via a communication network. This communication network can be a wired communication network or a wireless communication network. The wired communication network can be a communication network based on power line carrier technology, and the wireless communication network can be a local area network (LAN) or a wide area network (WAN). The LAN can be a Wi-Fi network, a Zigbee network, a mobile communication network, or a satellite communication network, etc.

[0055] like Figure 1 As shown, the user device 110 may include a computer, mobile phone, or information processing center, etc., as a smart terminal. The user device 110 can act as the initiator of model training operations or expression-driven parameter determination operations, sending requests to the computing device 120. The computing device 120 can be a cloud server, network server, application server, or management server, etc., with data processing capabilities, used to implement training and generation methods. The server can be configured with a deep learning processor, which can be a single-core deep learning processor (DLP-S) neuron or a multi-core deep learning processor (DLP-M). DLP-M is a multi-core extension based on DLP-S, interconnecting multiple DLP-S processors through Network-on-chip (Noc) protocols for inter-core communication, multicasting, and inter-core synchronization to complete deep learning tasks and question recommendation tasks.

[0056] like Figure 1 As shown, the aforementioned data storage system 130 can be a general term, including local storage and a database storing historical data. The database can be on the computing device 120, on other network servers, or on the data storage system 130. The data storage system 130 can be separate from the computing device 120 or integrated within the computing device 120.

[0057] In practical applications, the aforementioned user equipment can upload completed test questions and related user information to the computing device via a communication network. This information is stored in a data storage system. Simultaneously, the user equipment can send a test question recommendation request to the computing device via the communication network. Upon receiving the request, the computing device can retrieve historical test questions from the data storage system, analyze the user's current mastery of the target knowledge point based on these historical questions, and then select recommended questions from the question bank based on this mastery level. The recommended questions are then fed back to the user equipment via the communication network. Of course, when sending a test question recommendation request to the computing device via the communication network, the user equipment can also send historical test questions to the computing device via the communication network, or respond to a request from the computer device by sending historical test questions via the communication network, analyzing the user's current mastery of the target knowledge point based on these historical questions.

[0058] The determination of the degree of dynamic knowledge mastery in the exemplary embodiments of this disclosure can be applied to servers or chips in servers. The method of the exemplary embodiments of this disclosure is described in detail below with reference to the accompanying drawings.

[0059] Figure 2 A schematic flowchart of a method for determining the dynamic level of knowledge mastery, an exemplary embodiment of the present disclosure, is shown. The method for determining the dynamic level of knowledge mastery, an exemplary embodiment of the present disclosure, includes:

[0060] Step 201: Determine the knowledge point codes for multiple questions answered by the same user. The knowledge point codes include the knowledge point description information of the questions and the answer results.

[0061] In practical applications, before determining the knowledge point encoding of multiple test questions based on multiple test questions answered by the same user, the exemplary embodiment of this disclosure determines the knowledge point description information of the test questions based on the knowledge point description parameters and weights of each layer of knowledge points involved in the knowledge graph.

[0062] For example, the knowledge point description parameters of this exemplary embodiment can be determined using a knowledge point description model, which may include a deep language representation subnetwork and a fully connected subnetwork to obtain test result description information. The architecture of the deep language representation subnetwork can be the same as that of the BERT model. The text of the test question can be input into the deep language representation subnetwork to obtain deep representation information of knowledge points. Then, the fully connected subnetwork is used to process the deep representation information of knowledge points to obtain knowledge point description parameters, which can also be called knowledge point labels or knowledge label depth vectors.

[0063] When constructing a knowledge graph, a question bank can be established. Based on the knowledge point structure within the questions, connections between different knowledge points are built to form the knowledge graph. For example, the question bank can be categorized based on knowledge point information, ensuring that each category of questions includes at least the same main knowledge points. On this basis, the knowledge points contained in each category of questions are extracted, and steps such as fusion, data model construction, and quality assessment are performed to obtain the knowledge graph. It should be understood that, in the exemplary embodiments of this disclosure, the main knowledge point refers to the knowledge point with the highest weight in the knowledge graph among the knowledge points contained in the questions; this knowledge point is then called the main knowledge point.

[0064] The weights of knowledge points at each level in this exemplary embodiment can be determined with reference to the knowledge graph level where the knowledge point resides. For example, a mapping relationship between the knowledge graph level where a knowledge point resides and the knowledge point weight can be designed. When it is necessary to query the weight of a knowledge point, the level corresponding to the knowledge point in the knowledge graph should first be determined, and then the weight of the knowledge point should be determined according to the mapping relationship. In the knowledge graph, each level has its corresponding knowledge point weight, and the knowledge points located in the same level of the knowledge graph have the same weight.

[0065] When answering test questions within a knowledge graph, the hierarchical parameters of the knowledge points are positively correlated with their descriptive parameters. The deeper the knowledge point's level, the greater its weight, meaning a larger hierarchical parameter. In this case, a larger knowledge point weight corresponds to a larger descriptive parameter, indicating more explicit descriptive information. Furthermore, by controlling the knowledge point's weight, one can control the descriptive parameters and thus the descriptive information.

[0066] For example, when the description parameters of a knowledge point are non-linearly related to its hierarchical parameters, the weight of that knowledge point can satisfy Equation 1:

[0067]

[0068] Among them, h i Let h represent the weight of the knowledge point at level i, where i represents the level to which the knowledge point belongs, and i is an integer greater than 1. As can be seen from Equation 1, the weight h of the knowledge point at level i... i It has a positive correlation with the level i to which the knowledge point belongs.

[0069] For example, the description information of each knowledge point is matched with the weighted sum of the knowledge point description parameters of each answer question in the knowledge graph. For instance, the knowledge point description parameters of each answer question in the knowledge graph can construct a set of knowledge point description parameters {k1, k2, ..., k}. n-1 k n}, where n represents the total number of layers in the knowledge graph involved in answering the test questions, and kn This represents the knowledge point description parameters located at the nth layer of the knowledge graph. A larger n indicates a deeper level of the knowledge point within the knowledge graph. Based on this, the set of knowledge point description parameters {k1, k2, ..., kn} involved in the test questions within the knowledge graph is used. n-1 k n By performing a weighted summation, we can obtain the knowledge point description information for this question. This knowledge point description information can be expressed as Equation 2:

[0070] kh=h1*k1+h2*k2+…+h n *k n Formula 2

[0071] Where h1 represents the weight of the knowledge point in the first layer, h2 represents the weight of the knowledge point in the second layer, and h... n Let kh represent the weight of the knowledge point at the nth level, kh be the knowledge point description information for the question, k1 be the knowledge point description parameter of the first level knowledge point, k2 be the knowledge point description parameter of the second level knowledge point, and k be the weight of the knowledge point at the nth level knowledge point. n-1 k is the knowledge point description parameter for the (n-1)th level knowledge point. n These are the knowledge point description parameters for the nth level of knowledge points.

[0072] For example, in the exemplary embodiments of this disclosure, the multiple test questions are answered at different times, and the multiple test questions contain the same knowledge points. A single test question may involve one or more knowledge points. When each test question includes multiple knowledge points, the multiple test questions involve the same knowledge points, but the main knowledge points involved in the multiple test questions may be the same or different. For example, test question A involves three knowledge points a, b, and c, and test question B also involves three knowledge points a, b, and c, but the main knowledge point of test question A is knowledge point a, and the main knowledge point of test question B is knowledge point b, but test question A and test question B involve the same knowledge points.

[0073] The answer results of the test questions in this exemplary embodiment include two possible outcomes: a correct answer and an incorrect answer. Therefore, since the knowledge point encoding includes the answer results, and user answer results may differ, the knowledge point encoding for the same knowledge point may be the same or different.

[0074] When a user first starts answering questions about knowledge point A, their error rate is relatively high. However, when a user repeatedly answers questions about knowledge point A over a long period, their error rate decreases, meaning their accuracy rate is higher. Therefore, if multiple questions are answered identically, the knowledge point codes determined based on answers given at different times may differ.

[0075] For example, let x be the knowledge point code for the question in the i-th time period. i ={q i ,a i}, q i The descriptive characteristics of the knowledge points in the i-th time period for answering the test questions; a i This represents the answer result of the test question in the i-th time period. When the answer result is correct, a i If the answer is incorrect, then a is 1. i The value is 0. Here, i is an integer greater than or equal to 1 and less than or equal to t. The larger i is, the later the time for answering the questions. Based on this, the knowledge point code x for answering the questions is... i It can not only reflect the user's knowledge point description information in the i-th time period, but also the answer results for the description features of that knowledge point.

[0076] For example, assuming t = 30, when i = 1, the knowledge point code of the question in the first answering period is x1 = {q1, a1}, where q1 is the description of the knowledge point of the question in the first answering period, and a1 represents the answer result of the question in the first period. a1 = 0 indicates that the early users had a relatively poor grasp of this question. As the number of answers increases, when i = 30, the knowledge point code of the question in the 30th answering period is x1 = {q1, a1}. 30 ={q 30 ,a 30}, q 30 For the knowledge point description information in the 30th answering time of the test questions, a 30 This indicates the answer result of the test questions in the 30th time slot, a 30 =1 indicates that after repeated attempts, the user has a relatively high level of mastery of the question.

[0077] As can be seen, in the exemplary embodiments of this disclosure, the level of mastery of the same user for the same test question changes as the number of times the test is answered.

[0078] Step 202: Determine the user's dynamic knowledge mastery level based on the knowledge point codes of multiple test questions. Since the test questions were answered at different times, the knowledge point codes include not only the knowledge point descriptions but also the answer results. Therefore, based on the knowledge point codes of multiple test questions, the changes in the answer results over time can be determined. The answer results of the test questions can indirectly reflect the user's dynamic knowledge mastery level for that question. Therefore, the user's dynamic knowledge mastery level can be obtained based on the changes in the answer results over time.

[0079] In practical applications, exemplary embodiments of this disclosure can input the knowledge point codes of multiple test questions into a deep knowledge tracing model. The deep knowledge tracing model can then obtain the user's dynamic knowledge mastery level based on these codes. Since the knowledge point codes of the test questions include the knowledge point description information and the answer results, the knowledge point description information of multiple test questions can be used as part of the deep knowledge tracing model. Combined with historical answer sequence data formed by the answer results of multiple test questions, latent variable representations for knowledge can be mined, thereby dynamically monitoring the user's state changes regarding knowledge points.

[0080] For example, the deep knowledge tracking model of this exemplary embodiment can be a trained bidirectional LSTM network model. The bidirectional LSTM network model can model the user's learning state at each time period, thereby obtaining the user's dynamic knowledge mastery level. It is evident that the bidirectional LSTM network model can be used to record the user's knowledge mastery, thereby enhancing the mining of the user's knowledge mastery ability.

[0081] For example, suppose a user answers different questions at different times, resulting in the questions and answers for each time period (which can be viewed as a question sequence). Based on the chronological order of the answering times, the knowledge point codes for these questions can be coded as a model input sequence. This model input sequence is then divided into a forward sequence S1 and a reverse sequence S2, where the forward sequence S1 = {x1, x2, ..., x...} T}, S2={x T ,x T-1 ,……,x1}, where x1 is the knowledge point code for the first question answered in the first answering time period, and x2 is the knowledge point code for the second question answered in the second answering time period, x T-1 To encode the knowledge points of the question answered in the (T-1)th time period, x T This is the knowledge point encoding for the Tth question answered during the Tth answering period, where T is the total number of times the user answers the question, and T is an integer greater than or equal to 2.

[0082] Based on this, the forward sequence S1 and the reverse sequence S2 are input into the bidirectional LSTM network model, so that the bidirectional LSTM network model can determine the forward dependency vector based on the forward sequence S1 and the reverse dependency vector based on the reverse sequence S2. Then, the forward dependency vector and the reverse dependency vector are concatenated to obtain the user's dynamic knowledge mastery level.

[0083] The network parameters of the bidirectional LSTM model in this exemplary embodiment are divided into two parts: forward network parameters and backward network parameters. The bidirectional LSTM model can use the forward network parameters to predict the impact of correct answers on a user's mastery of a knowledge point as practice deepens, and use the backward network parameters to predict the impact of incorrect answers on a user's mastery of a knowledge point as the connections between knowledge points decrease. Therefore, this exemplary embodiment can fuse the forward and backward network parameters of the bidirectional LSTM model to model the user's knowledge mastery using the user's model input sequence, enabling the bidirectional LSTM model to output the user's dynamic level of knowledge mastery.

[0084] For example, the dynamic knowledge point mastery matrix of this exemplary embodiment may include knowledge point mastery parameters for a user on multiple knowledge points at different time periods, and the knowledge point mastery parameters may be the scores for mastering the knowledge point. It should be understood that the user's dynamic knowledge mastery level can be represented in the form of a dynamic knowledge point mastery matrix, which may be an M×N dynamic knowledge point mastery matrix. M represents the number of rows in the dynamic knowledge point mastery matrix, also known as the knowledge level vector dimension, which may be the total number of test questions answered in the time series, and N represents the number of columns in the dynamic knowledge point mastery matrix, which may be the total number of knowledge points.

[0085] As can be seen, the exemplary embodiments of this disclosure can dynamically diagnose the user's knowledge status and ability level based on the user's answer sequence, thereby making fuller use of deep networks to mine user answer data in time series. It has made improvements by taking the question knowledge tag vector as input and providing the user's dynamic knowledge mastery level through a deep knowledge tracking model, that is, the change of the user's knowledge status and ability level over time.

[0086] The test question recommendation method of this exemplary embodiment can be applied to a server or a chip in a server. The method of this exemplary embodiment is described in detail below with reference to the accompanying drawings.

[0087] Figure 3 A schematic flowchart of a test item recommendation method according to an exemplary embodiment of this disclosure is shown. The test item recommendation method according to an exemplary embodiment of this disclosure includes:

[0088] Step 301: Determine the user's current level of mastery of the target knowledge point based on the user's dynamic level of knowledge mastery. This level of dynamic knowledge mastery is determined by the knowledge dynamic mastery method described in the exemplary embodiments of this disclosure.

[0089] In practical applications, for an M×N dynamic mastery matrix of knowledge points, for a knowledge point in the p-th list, the user's current mastery level for a knowledge point in the p-th column can be represented by the dynamic mastery level of the knowledge point in the M-th row of the p-th column, where p is an integer greater than or equal to 1 and less than or equal to N. Therefore, the method of this exemplary embodiment can obtain the user's latest mastery level for a particular knowledge point, i.e., the current mastery level, from the dynamic mastery level, thus providing a basis for subsequent test question recommendations, making the recommended test questions more suitable for the user's current learning situation.

[0090] Step 302: Determine multiple candidate test questions containing the target knowledge point based on the user's current mastery level of the target knowledge point. It should be understood that the target knowledge point can be a single knowledge point or multiple knowledge points. Based on this, the test questions in the test bank can be classified according to a single knowledge point or according to a combination of multiple knowledge points.

[0091] An exemplary embodiment of this disclosure can employ the knowledge dynamic mastery determination method of this exemplary embodiment to determine the user's mastery level for each question in the question bank. After determining the current mastery level of a target knowledge point, from the multiple questions under that target knowledge point in the established question bank, questions that match the current mastery level of the target knowledge point are selected as candidate questions, thereby achieving the purpose of recalling questions.

[0092] In practical applications, a user's current level of mastery of the target knowledge point indicates that they can answer more difficult questions containing that knowledge point. Therefore, the difficulty of each candidate question can be positively correlated with the user's current level of mastery of the target knowledge point. Based on this, while classifying the questions in the question bank according to the number of knowledge points, a difficulty score can be assigned to each question to obtain a question difficulty rating. This rating can then be saved as a question attribute, which can include not only the difficulty rating but also the knowledge point, question structure, etc.

[0093] For example, the exemplary embodiment of this disclosure can also establish a mapping relationship between the difficulty score of the test questions and the dynamic mastery of knowledge. Based on the user's current mastery of the target knowledge point, the difficulty of the test questions corresponding to the current mastery level can be queried from the mapping relationship. Then, test questions with similar difficulty (difficulty difference of less than 5%) under the target knowledge point can be selected as candidate test questions to form a candidate test question set.

[0094] As can be seen, the exemplary embodiments of this disclosure can find test questions that match the current mastery level of the target knowledge point based on the mapping relationship between the test question difficulty score and the dynamic mastery level of knowledge, thereby ensuring that the difficulty of the candidate test questions is appropriate for the user and avoiding the user answering test questions that are too easy or too difficult.

[0095] Step 303: Obtain at least one recommended question from multiple candidate questions containing the target knowledge point. Each candidate question may contain one or more knowledge points. When each candidate question contains multiple knowledge points, it has a main knowledge point, which can be the target knowledge point or a non-target knowledge point. Therefore, recommended questions with the main knowledge point limited to the target knowledge point can be obtained from multiple candidate questions containing the target knowledge point, so that the main knowledge point of each recommended question is the target knowledge point.

[0096] In practical applications, obtaining multiple target questions from multiple candidate questions containing target knowledge points can include: obtaining multiple selected questions from multiple candidate questions based on the question structure of the user's historical answers, and obtaining at least one recommended question with the same main knowledge point from multiple selected questions.

[0097] Considering that users tend to develop a habit of answering questions with similar or identical structures that target a particular knowledge point, it's apparent that they have a good grasp of the knowledge involved in that type of question. However, when the question structure changes, the user may not be able to effectively address the same knowledge point in a different question. Therefore, when multiple selected questions are derived from a pool of candidate questions based on the user's historical answers, each selected question has a different structure than the user's historical answers. This ensures that the recommended questions have a novel structure, thereby comprehensively improving the user's ability to answer questions targeting specific knowledge points.

[0098] For example, in this exemplary embodiment of the disclosure, the user's historical answers are those with correct answers. That is, user's historical answers with correct answers can be selected from the user's historical answers, and then selected as featured questions from those. When multiple featured questions are obtained from multiple candidate questions based on the question structure of user's historical answers, the question structure of user's historical answers with incorrect answers can be selected from the candidate questions, allowing users to repeatedly practice those user's historical answers with incorrect answers and avoiding unnecessary omissions of recommended questions.

[0099] For example, you can set question attributes for users' historical answers and candidate questions. These attributes can include question structure and answer results. The question structure can be defined by question type, answering method, etc. For instance, if users' historical answers are multiple-choice questions, while candidate questions are fill-in-the-blank questions, then their question structures can be considered different. As another example: if the answering method for users' historical answers is to check the answer box, while the answering method for candidate questions is to fill in the reply, then their question structures can also be considered different.

[0100] When multiple selected questions are obtained from the candidate question set based on the question structure of the user's historical answers, the system can select the user's historical answers with correct answers from the historical answers. Then, based on the question structure of the user's historical answers, the system can select candidate questions with different question structures from the multiple candidate questions as selected questions, thereby removing candidate questions with too high similarity and preventing users from repeatedly answering questions with similar structures.

[0101] For example, in an exemplary embodiment of this disclosure, the similarity between the question structure description information of the user's historical answers and the question structure description information of the selected questions is less than a first similarity threshold, which can be cosine similarity, etc.

[0102] In practical applications, a similarity comparison model can be used to compare the question structure descriptions of users' historical answers with those of candidate questions. Questions with high similarity are filtered out, and questions with similarity less than a first similarity threshold are retained. The filtering method can be to remove candidate questions with similarity greater than or equal to the first similarity threshold after the similarity comparison model outputs the actual similarity, and retain candidate questions with similarity less than the first similarity threshold as selected questions.

[0103] The test result description information of the exemplary embodiments of this disclosure may include information splicing features of each test question segment. It may be based on the information contained in the test question to split the test question into at least two test question segments, and perform in-depth analysis on the content contained in each test question segment before splicing them together, thereby comprehensively mining the test question content.

[0104] An exemplary embodiment of this disclosure can employ a test item structure description model to obtain test item structure description information. This test item structure description model may include a deep language representation subnetwork and a fully connected subnetwork to obtain test item result description information. The architecture of the deep language representation subnetwork can be the same as the BERT model.

[0105] For any text question, at least the question fragment and the stem fragment can be separated. Based on this, the question content can be divided into at least two text fragments according to the question structure. The at least two text fragments are preprocessed into short texts. Then, the at least two short texts are input into a deep language representation subnetwork to obtain deep text representation vectors of the at least two short texts. Then, the deep text representation vectors of the at least two short texts are concatenated through a fully connected subnetwork to obtain the question structure description information of the text question.

[0106] For example, when the test question is a text-based question, the process first retrieves the text-based test question to be split from the test question bank. Then, the text and data in the text-based test question to be split are labeled to obtain the text coordinate features. The text coordinate features and data coordinate features of the text-based test question to be split are then input into the test question splitting model to obtain the test question fragment D. 11 Question stem fragment Q 21 and test question option fragments O 31 This leads to the test question fragment D. 11 Question stem fragment Q 21 And the question options fragment O 31 Preprocessing into short text fragments of test questions (D) 12 The question stem is a short text Q. 22 And short text snippets of test question options O 32 Based on this, the short text D containing the test question fragments... 12 The question stem is a short text Q. 22 And short text snippets of test question options O 32 Inputting the data into the BERT model yields the deep text representation vector D corresponding to the short text fragment of the test question. 13 The deep text representation vector Q corresponding to the short text excerpt in the question stem. 23 And the deep text representation vector O corresponding to the short text of the test question options. 33 Next, the deep text representation vector D corresponding to the short text of the test question is... 13 The deep text representation vector Q corresponding to the short text excerpt in the question stem. 23 And the deep text representation vector O corresponding to the short text of the test question options. 33 By concatenating the information using a fully connected neural network, the structural description information r of this text-based test question can be obtained. i =(D 13 Q 23 O 33 ), where σ is the activation function in a fully connected neural network. It should be understood that when splitting test questions based on the question splitting model, the splitting is performed according to the question type, and the split question segments can be pre-set, depending on the actual situation.

[0107] In order to provide effective practice on target knowledge points, the target knowledge points in the exemplary embodiments of this disclosure can be the main knowledge points of the recommended test questions, thereby ensuring that the recommended test questions mainly test the user's mastery of the target knowledge points, or mainly improve the user's learning ability on the target knowledge points.

[0108] For example, the attributes of the selected test questions can be enriched to include the main knowledge point category and even the proportion of the main knowledge point. Based on this, recommended test questions with the main knowledge point category as the target knowledge point can be obtained from multiple selected test questions. Of course, since the deeper the knowledge point level, the greater its proportion in the test questions, the main knowledge point of each selected test question can also be obtained based on its level in the knowledge graph. Then, it can be determined whether this main knowledge point is the target knowledge point; if so, it is a recommended test question.

[0109] In some alternative approaches, when multiple recommended questions are obtained, these questions can be finely sorted, and a final question recommendation order can be established based on these sorting rules. An exemplary embodiment of this disclosure can recommend questions in the following recommendation order:

[0110] The first method involves prioritizing multiple recommended practice questions based on their difficulty level. For example, the difficulty level of recommended questions can be determined by their attributes, and then multiple questions can be recommended according to their difficulty priority. This difficulty priority can be from highest to lowest or lowest to highest. When recommended questions are presented in ascending order of difficulty, the resulting learning path is adopted by the user. The difficulty level of the recommended questions matches the user's current understanding of the target knowledge point, allowing the user to prioritize questions closest to their current level of knowledge, thus increasing their learning motivation.

[0111] The second approach involves prioritizing multiple recommended practice questions based on their proportion of key knowledge points. For example, the proportion of key knowledge points in each recommended question can be obtained from its attributes. Then, multiple recommended questions are recommended according to this priority. It should be understood that this proportion can be from highest to lowest or vice versa. When recommended questions are presented in descending order of key knowledge point proportions, the resulting learning path is more readily adopted by the user. This ensures that the first recommended questions have the highest proportion of target knowledge points, thereby improving the user's learning efficiency for those target knowledge points.

[0112] The third method involves ranking multiple recommended test questions according to the similarity priority between the knowledge point descriptions of the recommended test questions and the target knowledge points. This similarity priority can be either from highest to lowest or lowest to highest.

[0113] In practical applications, a knowledge point description model can be used to determine the description parameters of each knowledge point in the recommended test questions. Then, the level of each main knowledge point in the knowledge graph corresponding to the recommended test questions can be determined. Based on the mapping relationship between the knowledge graph level of each knowledge point and the knowledge point weight, the corresponding knowledge point weight can be retrieved. Then, based on each knowledge point description parameter and the corresponding knowledge point weight, the knowledge point description information of the recommended test questions can be obtained. On this basis, the similarity between the knowledge point description information of the recommended test questions and the target knowledge points can be compared; the similarity can be cosine similarity, etc. It should be understood that the knowledge point description parameters and knowledge point weights of the exemplary embodiments of this disclosure can be referred to the preceding text and will not be repeated here.

[0114] When recommended test questions are presented in descending order of similarity, the resulting learning path is adopted by the user. This ensures that the target knowledge points are maximized in the first recommended test questions presented to the user, thereby improving the user's learning efficiency for the target knowledge points.

[0115] In one optional embodiment, the recommended test questions in this exemplary embodiment are multiple, and the recommended test questions can be further refined to avoid similar recommended test questions. The method in this exemplary embodiment may further include: comparing the similarity of the deep complete description information of two recommended test questions; if the similarity of the deep complete description information is greater than or equal to a second similarity threshold, deleting one of the two recommended test questions. In this case, the recommended test questions recommended to the user do not contain recommended test questions with excessively high similarity, which can ensure that the user can effectively practice the learning ability of the target knowledge points through recommended test questions while avoiding unnecessary repetitive practice, thereby improving learning efficiency.

[0116] For example, the deep complete description information of the exemplary embodiments of this disclosure can integrate the knowledge point information and question content information of the recommended test questions. Based on this, the deep complete description information of the exemplary embodiments of this disclosure can include the knowledge point description features of the recommended test questions and the question structure description information of the recommended test questions. It should be understood that the second similarity threshold can be set according to actual circumstances, and it can be cosine similarity.

[0117] Figure 4 A schematic diagram illustrating the process for determining deep and complete descriptive information in an exemplary embodiment of this disclosure is shown. Figure 4As shown, the knowledge point description model 401 can be used to obtain the description parameters of each knowledge point of the recommended test questions. Then, a weighted fusion method is used to fuse the description parameters of each knowledge point to obtain the knowledge point description features. The test question structure description model 402 can be used to obtain the test question structure description information of the recommended test questions. Then, a feature fusion network 403 (such as a fully connected network) can be used to concatenate the description features of the knowledge points and the description information of the test question structure to obtain the deep and complete description information of the recommended test questions. Therefore, the deep and complete description information of the recommended test questions in the exemplary embodiment of this disclosure can more comprehensively and accurately evaluate the recommended test questions from the depth information of the test question structure and the depth information of the knowledge points, thereby improving the accuracy of similarity comparison and ensuring that the recommended test questions can comprehensively, accurately and efficiently improve the user's learning ability. It should be understood that the process of obtaining the description parameters of each knowledge point and the description information of the test question structure in the exemplary embodiment of this disclosure can be referred to the above, and will not be repeated here.

[0118] As can be seen from the test question recommendation method of the exemplary embodiments of this disclosure: the exemplary embodiments of this disclosure obtain the user's current mastery of the target knowledge point based on the user's dynamic mastery of knowledge, and then combine information such as test question difficulty, test question knowledge point description features, and in-depth and complete description information to filter, accurately sort, and match the test questions in the test question bank, and select recommended questions with high suitability to recommend to the user.

[0119] One or more technical solutions provided in the exemplary embodiments of this disclosure determine the knowledge point codes for corresponding test questions based on multiple test questions answered by the same user. Since the multiple test questions were answered at different times and contain the same knowledge points, the knowledge point codes of the multiple test questions can be used to obtain dynamic change information of the knowledge point-related content. Furthermore, since the knowledge point code includes the knowledge point description information and the answer result of the test question, and the answer result reflects the user's level of mastery of the knowledge point, the user's dynamic level of knowledge mastery can be determined based on the knowledge point codes of the multiple test questions.

[0120] Based on this, when applying the method of the exemplary embodiment of this disclosure to test question recommendation, it can obtain the user's current mastery level of the target knowledge point from the user's dynamic knowledge mastery level, and then select test questions containing the target knowledge point and matching the current mastery level from the test question bank as candidate test questions, and then filter recommended test questions from the candidate test questions. Based on this, when recommending test questions to users, the recommended test questions can be adapted to the user's current mastery level of the target knowledge point, ensuring that the recommended test questions match the user's learning ability, making the recommended test questions more suitable for the user's current learning situation and improving the user's learning efficiency for the target knowledge point. At the same time, since the main knowledge point of the recommended test questions is the target knowledge point, it can ensure that the recommended test questions can primarily provide users with ability improvement services for the target knowledge point. Therefore, the method of the exemplary embodiment of this disclosure can improve the user's learning efficiency for the target knowledge point.

[0121] In summary, the exemplary embodiments of this disclosure fully utilize the powerful data mining and representation capabilities of deep neural networks. By using deep information to describe knowledge points and question structures, it uncovers the user's mastery of the knowledge points contained in the questions, accurately describing the question structure and the user's knowledge mastery, thereby comprehensively and meticulously assessing the user's dynamic knowledge mastery. Based on this, various filtering methods, such as candidate question selection and main knowledge point selection, which are highly adapted to the user's current mastery of the target knowledge points, are combined with the characteristics of the questions themselves. This saves the user's time, provides the most suitable questions, and thus quickly and accurately improves the user's learning ability and efficiency.

[0122] The foregoing primarily describes the solutions provided by the embodiments of this disclosure from the perspective of the server. It is understood that, in order to implement the above functions, the server includes the corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein, this disclosure can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0123] This disclosure embodiment can divide the server into functional units according to the above method example. For example, it can divide each function into a separate functional module, or it can integrate two or more functions into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this disclosure embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0124] In the case of dividing each functional module according to its corresponding functions, an exemplary embodiment of this disclosure provides a device for determining the dynamic level of knowledge mastery, which can be a server or a chip applied to a server. Figure 5 A schematic block diagram of the functional modules of a knowledge dynamic mastery level determination device according to an exemplary embodiment of the present disclosure is shown. Figure 5 As shown, the knowledge dynamic mastery level determination device 500 includes:

[0125] A device for determining the degree of dynamic knowledge mastery, characterized in that the device comprises:

[0126] The encoding module 501 is used to determine the knowledge point codes of multiple test questions answered by the same user. The multiple test questions are answered at different times. The knowledge point codes include the knowledge point description information and the answer results of the test questions. The multiple test questions contain the same knowledge points.

[0127] The determination module 502 is used to determine the dynamic mastery of knowledge based on the knowledge point codes of multiple answer questions.

[0128] In one possible implementation, the determining module 502 is further configured to, before determining the knowledge point encoding of multiple answer questions based on multiple answer questions from the same user, determine the knowledge point description information of each answer question based on the knowledge point description parameters and weights of each layer of knowledge points involved in the knowledge graph.

[0129] In one possible implementation, the hierarchical parameters of the knowledge points involved in the answer question in the knowledge graph are positively correlated with the description parameters of the knowledge points.

[0130] In one possible implementation, the knowledge point description parameters are non-linearly related to the hierarchical parameters of the knowledge point.

[0131] In one possible implementation, the knowledge point description information is matched with the weighted sum of the knowledge point description parameters of each layer involved in the knowledge graph of the answer question.

[0132] In one possible implementation, the user's dynamic knowledge mastery level includes parameters on the user's mastery of multiple knowledge points at different time periods.

[0133] In the case of dividing each functional module according to its corresponding function, an exemplary embodiment of this disclosure provides a test question recommendation device, which can be a server or a chip applied to a server. Figure 6 A schematic block diagram of the functional modules of a test item recommendation device according to an exemplary embodiment of the present disclosure is shown. Figure 6 As shown, the test question recommendation device 600 includes:

[0134] The determining module 601 is used to determine the user's current mastery level of the target knowledge point based on the user's dynamic mastery level of knowledge, and to determine multiple candidate test questions containing the target knowledge point based on the user's current mastery level of the target knowledge point. The dynamic mastery level of knowledge is determined by the method described in the exemplary embodiment of this disclosure.

[0135] The filtering module 602 is used to obtain at least one recommended question from multiple candidate questions containing target knowledge points, wherein the main knowledge point of each recommended question is the target knowledge point.

[0136] In one possible implementation, the difficulty of each candidate question is positively correlated with the user's current level of mastery of the target knowledge point.

[0137] In one possible implementation, the determining module is further configured to obtain multiple selected questions from multiple candidate questions based on the question structure of the user's historical answers, and to obtain at least one recommended question from the multiple selected questions whose main knowledge point is the target knowledge point, wherein the question structure of each selected question is different from the question structure of the user's historical answers.

[0138] In one possible implementation, the similarity between the question structure description information of the user's historical answers and the question structure description information of the selected questions is less than a similarity threshold, and the question structure description information includes the information splicing features of each question segment; and / or,

[0139] The user's historical test questions are those with correct answers.

[0140] In one possible implementation, the number of recommended test questions is multiple, and the recommendation order of the multiple recommended test questions is as follows:

[0141] The recommended test questions are prioritized according to their difficulty; or,

[0142] The recommended test questions are ordered by the priority of the main knowledge points; or,

[0143] The priority order of similarity between the knowledge point description information of the recommended test questions and the target knowledge points.

[0144] In one possible implementation, the number of recommended test questions is multiple, and the filtering module 603 is further used to compare the similarity of the deep complete description information of two recommended test questions. If the similarity of the deep complete description information is greater than or equal to a second similarity threshold, one of the two recommended test questions is deleted. The deep complete description information includes the knowledge point code of the recommended test question and the test question structure description information of the recommended test question.

[0145] Figure 7 A schematic block diagram of a chip according to an exemplary embodiment of the present disclosure is shown. Figure 7 As shown, the chip 700 includes one or more (including two) processors 701 and a communication interface 702. The communication interface 702 can support the server in performing the data transmission and reception steps in the above method, and the processor 701 can support the server in performing the data processing steps in the above method.

[0146] Optional, such as Figure 7 As shown, the chip 700 also includes a memory 703, which may include read-only memory and random access memory, and provides operation instructions and data to the processor. A portion of the memory may also include non-volatile random access memory (NVRAM).

[0147] In some implementations, such as Figure 7 As shown, processor 701 executes corresponding operations by calling operation instructions stored in memory (which may be stored in the operating system). Processor 701 controls the processing operations of any terminal device; processor can also be called a central processing unit (CPU). Memory 703 may include read-only memory and random access memory, and provides instructions and data to processor 701. A portion of memory 703 may also include NVRAM. For example, in applications, memory, communication interfaces, and other components are coupled together via a bus system, which may include, in addition to a data bus, a power bus, a control bus, and a status signal bus, etc. However, for clarity, in... Figure 7 The general designated all buses as Bus System 704.

[0148] The methods disclosed in the embodiments of this disclosure can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above methods can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an ASIC, a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above methods.

[0149] Exemplary embodiments of this disclosure also provide an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to cause the electronic device to perform a method according to an embodiment of this disclosure.

[0150] Exemplary embodiments of this disclosure also provide a non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform a method according to embodiments of this disclosure.

[0151] Exemplary embodiments of this disclosure also provide a computer program product, including a computer program, wherein, when executed by a processor of a computer, the computer program is used to cause the computer to perform a method according to an embodiment of this disclosure.

[0152] refer to Figure 8The present invention describes a structural block diagram of an electronic device that can serve as a server or client of the present disclosure, which is an example of a hardware device that can be applied to various aspects of the present disclosure. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0153] like Figure 8 As shown, the electronic device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. The RAM 803 may also store various programs and data required for the operation of the device 800. The computing unit 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0154] Multiple components in electronic device 800 are connected to I / O interface 805, including: input unit 806, output unit 807, storage unit 808, and communication unit 809. Input unit 806 can be any type of device capable of inputting information to electronic device 800. Input unit 806 can receive input digital or character information and generate key signal inputs related to user settings and / or function control of electronic device. Output unit 807 can be any type of device capable of presenting information and may include, but is not limited to, a display, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 804 may include, but is not limited to, disk and optical disk. Communication unit 809 allows electronic device 800 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers, and / or chipsets, such as Bluetooth™ devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.

[0155] The computing unit 801 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above. For example, in some embodiments, the methods of exemplary embodiments of this disclosure can be implemented as computer software programs tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 800 via ROM 802 and / or communication unit 809. In some embodiments, the computing unit 801 can be configured to perform the methods of exemplary embodiments of this disclosure by any other suitable means (e.g., by means of firmware).

[0156] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0157] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0158] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0159] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0160] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

[0161] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this disclosure are performed, in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a terminal, a user equipment, or other programmable device. The computer program or instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center integrating one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; it can also be an optical medium, such as a digital video disc (DVD); or it can be a semiconductor medium, such as a solid-state drive (SSD).

[0162] Although this disclosure has been described in conjunction with specific features and embodiments, it will be apparent that various modifications and combinations can be made therein without departing from the spirit and scope of this disclosure. Accordingly, this specification and drawings are merely exemplary illustrations of the disclosure as defined by the appended claims and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this disclosure. It is obvious that those skilled in the art can make various alterations and modifications to this disclosure without departing from its spirit and scope. Thus, this disclosure is also intended to include any such modifications and modifications that fall within the scope of the claims of this disclosure and their equivalents.

Claims

1. A method for determining the degree of dynamic knowledge mastery, characterized in that, The method includes: The knowledge point codes for multiple test questions answered by the same user are determined, wherein the multiple test questions are answered at different times, and the knowledge point codes include the knowledge point description information and the answer results of the test questions, and the multiple test questions contain the same knowledge points; wherein, the knowledge point code is a vector containing the knowledge point description information and the answer results, and the knowledge point description information is determined for each test question based on the knowledge point description parameters and the weights of each layer of knowledge points involved in the knowledge graph; The knowledge point codes of multiple test questions are arranged in chronological order according to the time period of the test. Based on the changes in the answer results of the arranged knowledge point codes over time, the user's dynamic mastery of knowledge is determined.

2. The method according to claim 1, characterized in that, The hierarchical parameters of the knowledge points involved in the test questions within the knowledge graph are positively correlated with the descriptive parameters of the knowledge points.

3. The method according to claim 1, characterized in that, The description parameters of the knowledge point are non-linearly related to the hierarchical parameters of the knowledge point.

4. The method according to any one of claims 1 to 3, characterized in that, The knowledge point description information is a weighted sum and match with the knowledge point description parameters of each layer involved in the knowledge graph of the answer question.

5. The method according to any one of claims 1 to 3, characterized in that, The user's dynamic knowledge mastery level includes the user's knowledge mastery parameters for multiple knowledge points at different time periods.

6. A test item recommendation method, characterized in that, The method includes: The user's current level of knowledge mastery for the target knowledge point is determined based on the user's dynamic level of knowledge mastery, wherein the user's dynamic level of knowledge mastery is determined by the method described in any one of claims 1 to 5; Based on the user's current level of mastery of the target knowledge point, multiple candidate test questions containing the target knowledge point are determined; At least one recommended question is obtained from multiple candidate questions containing the target knowledge point. The main knowledge point of each recommended question is the target knowledge point, wherein the main knowledge point is the knowledge point with the highest weight among the multiple knowledge points included in the recommended question.

7. The method according to claim 6, characterized in that, The difficulty of each candidate question is positively correlated with the user's current level of mastery of the target knowledge point.

8. The method according to claim 6, characterized in that, The step of obtaining at least one recommended question from multiple candidate questions containing the target knowledge point includes: Based on the question structure of the user's historical answers, multiple selected questions are obtained from multiple candidate questions, and the question structure of each selected question is different from the question structure of the user's historical answers. At least one recommended question is obtained from the selected questions, with the main knowledge point being the target knowledge point.

9. The method according to claim 8, characterized in that, The similarity between the question structure description information of the user's historical answers and the question structure description information of the selected questions is less than a first similarity threshold, and the question structure description information includes the information splicing features of each question segment; and / or, The user's historical test questions are those with correct answers.

10. The method according to any one of claims 6 to 9, characterized in that, The number of recommended test questions is multiple, and the recommended order of the multiple recommended test questions includes: The recommended test questions are prioritized according to their difficulty; or, The recommended test questions are ordered by the priority of the main knowledge points; or, The priority order of similarity between the knowledge point description information of the recommended test questions and the target knowledge points.

11. The method according to any one of claims 6 to 9, characterized in that, The number of recommended test questions is multiple, and the method further includes: Compare the similarity of the in-depth complete description information of the two recommended test questions. If the similarity of the in-depth complete description information is greater than or equal to the second similarity threshold, delete one of the two recommended test questions. The in-depth complete description information includes the knowledge point encoding of the recommended test question and the test question structure description information of the recommended test question.

12. A device for determining the degree of dynamic knowledge mastery, characterized in that, The device includes: An encoding module, configured to determine the knowledge point encoding of the corresponding answered test questions based on multiple answered test questions of the same user. The answering time periods of the multiple answered test questions are different. The knowledge point encoding includes the knowledge point description information and the answering result of the answered test question, and the knowledge points included in the multiple answered test questions are the same. Among them, the knowledge point encoding is a vector including the knowledge point description information and the answering result. The knowledge point description information is determined by, for each answered test question, based on the knowledge point description parameters and knowledge point weights of each layer involved in the answered test question in the knowledge graph.

13. A test question recommendation device, characterized in that, A determination module, configured to arrange the knowledge point encodings of the multiple answered test questions in the chronological order of the answering time periods, and determine the user's dynamic knowledge mastery degree based on the change of the answering results in the arranged knowledge point encodings over time. The device includes: A determination module, configured to determine the current mastery degree of the user for the target knowledge point based on the user's dynamic knowledge mastery degree, and determine multiple candidate test questions containing the target knowledge point based on the current mastery degree of the user for the target knowledge point. The user's dynamic knowledge mastery degree is determined by the method according to any one of claims 1 to 5.

14. An electronic device, characterized in that, A recommendation module, configured to obtain at least one recommended test question from the multiple candidate test questions containing the target knowledge point. The main knowledge point of each recommended test question is the target knowledge point. Among them, the main knowledge point is the knowledge point with the largest weight among the multiple knowledge points included in the recommended test question. Includes: A processor; And, A memory storing a program; 15. A non-transitory computer-readable storage medium, characterized in that, Among them, the program includes instructions, and when the instructions are executed by the processor, the processor is caused to execute the method according to any one of claims 1 to 11. The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 11.

Images